主题:Mitigating congestions among robots: Joint destination and parcel- to-grid assignment problem in robotic sorting systems

题目： 缓解机器人拥堵：机器人分拣系统中的地址与包裹到格口的联合分配问题

时间：2026年1月6日下午15:30-16:30

地点：管理科研楼第一教室

主讲人： Zhensheng Zhou, University of Science and Technology of China

Bio: Zhensheng Zhou received his Ph.D. in Management Science and Engineering from University of Science and Technology of China in November 2025. His main research interests are logistics and warehousing. He has papers in EJOR, and revision at MSOM.

Abstract: 

Robotic sorting systems (RSSs) are a novel type of sorting system that uses hundreds of robots to sort parcels by destination. A major concern in RSSs is congestions among robots, which incur unproductive waiting of robots and affect RSS's sorting efficiency. To mitigate congestions, this paper studies the assignment of destinations and parcels to grids. We first identify factors that significantly affect RSS's efficiency using descriptive and regression analysis. We then build a two-stage integer programming model that incorporates those factors' impact. For the first stage model, we propose a provably near-optimal -approximation algorithm, where  is a function of the parcel quantity and RSS's layout-related parameters. While for the second stage model, a decomposition-based heuristic is developed and its near optimality is numerically validated by a LP-relaxation-based lower bound model. Simulation results show that our method can achieve 29% higher throughput and 21% lower congestion compared with case companies' methods. Our assignment solution has some important managerial implications. First, throughput of RSSs that use our assignment solution is fairly insensitive to RSS's layouts (or to parcel quantity vectors) in a wide range around the optimal layout (or the initial vector). This suggests that a near optimal throughput can be achieved as long as the layout of RSS in use (or a roughly estimated parcel quantity vector utilized by our method for generating a solution) is within the abovementioned wide range. Second, a solution with "absolute balance of parcels among grids" (companies' policy) does not mean largest throughput. This is due to spatial heterogeneity of multiple grids assigned to the same destination and the trade-off between workload balance and total travel distance caused by this heterogeneity.